| Comment One YouTube comment or reply, fully resolved: the comment text, its exact YouTube comment id, the thread it belongs to and its reply depth, like count with an explicit approximate flag, reply count, the author's display name, channel id, channel URL, avatar and verified/creator flags, whether the creator hearted it, YouTube's own published-time string and an approximate ISO timestamp, plus the video id, URL, title and channel it came from. Deduplicated on YouTube's comment id, so a comment reached through two different inputs, or seen again on a scheduled re-run, is charged once. Videos with comments disabled, videos that do not exist, per-video rows, run summaries, the coverage report and every error row are free, and there is no run-start fee. | | |
| Enrichment One comment scored by the built-in analysis layer: sentiment label and a normalised -1..1 score, a toxicity triage score, extracted content keywords, the dominant writing system, whether the comment asks a question, and any video timestamps it cites. This is a lexicon signal built for triage and sorting at volume, not a moderation verdict, and no accuracy figure is claimed for it. Set enrichComments to false and this event never fires; it is charged only for comments actually scored, and never for free rows. | | |
| Video checked, no comment billed One video whose comments were actually read but that billed no comment: a monitoring re-run where nothing has been posted since your last run, or a video whose comments were all removed by your filters. Charged once per video per run, never on a video that billed a comment, and never on one that was private, deleted, has comments disabled or failed to load - those stay free and return an error row. This is what keeps a monitoring schedule honest: a quiet video costs one check instead of nothing, and a busy one is billed by comment as before. | | |